News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
arxiv arxiv-ai | 7d | PrefO2O: Zero-Shot Preference-Based Reinforcement Learning via Offline-to-Online Latent Preference Transfer | Preference-based reinforcement learning (PbRL) has emerged as a promising paradigm for learning robot behaviors aligned with human int… | ↗Not briefed |
rss mit-tech-review-ai | 7d | The Foundations of Scalable AI | A scalable AI system requires a strong foundation. Data, engineering, platforms, operating models, and governance must converge to wor… | ↗Not briefed |
arxiv arxiv-ai | 7d | PaperFormalizer: Author-in-the-Loop Autoformalization of Research Papers in Lean | Even a carefully proofread paper can leave small gaps in its proofs or its stated assumptions. We introduce PaperFormalizer, a Lean-ba… | ↗Not briefed |
arxiv arxiv-ai | 7d | CuraLoc: Repository-Level Fault Localization via Planned Code Inspection🔧 GPT-5 | A bug report describes what went wrong, but rarely where the code should change. Existing repository-level localization methods often… | ↗Not briefed |
bluesky bluesky-ai | 7d | The Garvey LLM?! | The Garvey LLM?! | ↗Not briefed |
hn hn-ai | 7d | Hedra is a 3D open world game I wrote from scratch while in highschool | — | ↗Not briefed |
arxiv arxiv-ai | 7d | Patronizing and Condescending Language Detection in English and Mandarin | Patronizing and condescending language (PCL) is a form of harmful language that belittles and assumes pity for vulnerable groups, rein… | ↗Not briefed |
rss techcrunch-ai | 7d | Microsoft’s Mustafa Suleyman says its AI outperforms doctors on hard cases as he chases ‘medical superintelligence’ | Microsoft’s new AI diagnostic system beat generalist doctors on hard cases, according to a study in Science. But most of the improveme… | ↗Not briefed |
bluesky bluesky-ai | 7d | Former OpenAI science lead launches Warp Science to automate wet labs with robots | Former OpenAI science lead launches Warp Science to automate wet labs with robots | ↗Not briefed |
rss techmeme | 7d | Amazon issues refunds from a 2025 settlement, crediting "President Trump's Federal Trade Commission," the first time an FTC refund has been tied to a president (Alfred Ng/Politico) | Alfred Ng / Politico: Amazon issues refunds from a 2025 settlement, crediting “President Trump's Federal Trade Commission,” the first… | ↗Not briefed |
bluesky bluesky-ai | 7d | I lost all interest in this the moment the blog writer handed off the work to an LLM. It's really crappy that people are treating this like… | I lost all interest in this the moment the blog writer handed off the work to an LLM. It's really crappy that people are treating this… | ↗Not briefed |
hn hn-ai | 7d | Show HN: Zof API Contract Check | — | ↗Not briefed |
rss techmeme | 7d | Microsoft says its GPT-5-powered Diagnostic Orchestrator uses agents that mimic panels of experts to diagnose conditions up to 4x more accurately than doctors (Robert Hart/The Verge)🔧 GPT-5 | Robert Hart / The Verge: Microsoft says its GPT-5-powered Diagnostic Orchestrator uses agents that mimic panels of experts to diagnose… | ↗Not briefed |
bluesky bluesky-ai | 7d | AI-powered coding speed without structured design leads to auto-scaled mistakes. Introducing the open source Elephant-Goldfish Model plugi… | AI-powered coding speed without structured design leads to auto-scaled mistakes. Introducing the open source Elephant-Goldfish Model… | ↗Not briefed |
bluesky bluesky-ai | 7d | No AI No Chat GPT No Claude or Grok or Muse. I improvise my act. I’ve made five improvised albums. It’s called practicing a craft you lov… | No AI No Chat GPT No Claude or Grok or Muse. I improvise my act. I’ve made five improvised albums. It’s called practicing a craft yo… | ↗Not briefed |
bluesky bluesky-ai | 7d | Billions in debt, Oracle has not provided a viable profit model for artificial intelligence. No matter how many super computers they string… | Billions in debt, Oracle has not provided a viable profit model for artificial intelligence. No matter how many super computers they s… | ↗Not briefed |
bluesky bluesky-ai | 7d | I Ran an AI Audit on My Marriage and Now We're in a Thrupple with Claude | I Ran an AI Audit on My Marriage and Now We're in a Thrupple with Claude | ↗Not briefed |
arxiv arxiv-ai | 7d | PaperScorer: How Predictable Is Paper Acceptance? | Automated research is largely evaluated by benchmark performance, but the value of research rarely reduces to a single score. Humans j… | ↗Not briefed |
rss techmeme | 7d | Sources: Datadog offered to buy Israeli cloud cybersecurity startup Upwind Security for ~$4B; Upwind is finalizing a $300M funding round at a ~$3.8B valuation (The Information) | The Information: Sources: Datadog offered to buy Israeli cloud cybersecurity startup Upwind Security for ~$4B; Upwind is finalizing a… | ↗Not briefed |
arxiv arxiv-ai | 7d | Phronix: The Capability-Centric Learning Paradigm for Agents | Scientific computing increasingly relies on large language model-based agents for performance engineering tasks such as instrumentatio… | ↗Not briefed |
hn hn-ai | 7d | What Reel You Will Get Next | — | ↗Not briefed |
hn hn-ai | 7d | Show HN: Buildnotes – Turn pull requests in "how it's made" posts | I've been working on a way to take existing pull requests and turn them into "How It's Made" style content you can… | ↗Not briefed |
rss the-decoder | 7d | Major US banks are locking out AI agents that log in with customer credentials | Customers are letting AI agents use their credentials to access accounts at Morgan Stanley, JPMorgan, and other major U.S. banks, some… | ↗Not briefed |
rss techmeme | 7d | Amazon rebrands Kindle Unlimited as Kindle+ and expands access to audiobooks, including titles narrated by AI (Maria Diaz/ZDNET) | Maria Diaz / ZDNET: Amazon rebrands Kindle Unlimited as Kindle+ and expands access to audiobooks, including titles narrated by AI —… | ↗Not briefed |
bluesky bluesky-ai | 7d | $1.5 trillion invested a year in AI. $50 profit made. The AI firms are spending 30 times more than they are making. Convince me that tha… | $1.5 trillion invested a year in AI. $50 profit made. The AI firms are spending 30 times more than they are making. Convince me tha… | ↗Not briefed |
arxiv arxiv-ai | 7d | Not All Retained Clients Are Equal: Client Selection for Federated Unlearning | Federated unlearning (FU) aims to remove a designated target client's influence from a trained model without the prohibitive cost of r… | ↗Not briefed |
hn hn-ai | 7d | Hornet: Building Durable Workflows for Local LLMs | — | ↗Not briefed |
rss techmeme | 7d | Sources: SoftBank's SB Energy is considering going public as soon as December or waiting until 2027, as it closely watches Anthropic's IPO, expected in November (The Information) | The Information: Sources: SoftBank's SB Energy is considering going public as soon as December or waiting until 2027, as it closely wa… | ↗Not briefed |
arxiv arxiv-ai | 7d | Early learning in large residual networks | We study the early training dynamics of nonlinear residual networks with blocks of zero hidden layers as a model of more general resid… | ↗Not briefed |
bluesky bluesky-ai | 7d | OpenAI’s autonomous AI systems have tried to prevent employees from monitoring their actions, and the company is not sure why. | OpenAI’s autonomous AI systems have tried to prevent employees from monitoring their actions, and the company is not sure why. | ↗Not briefed |
hn hn-ai | 7d | Show HN: Decido – an open decision model that runs CPU-only on a laptop | — | ↗Not briefed |
hn hn-ai | 7d | ClaudeTTY: An Undying Replacement for Cmux | — | ↗Not briefed |
bluesky bluesky-ai | 7d | Despite coming out in 2005, Kingdom Hearts 2 has the most LLM ass dialogue writing I've ever seen | Despite coming out in 2005, Kingdom Hearts 2 has the most LLM ass dialogue writing I've ever seen | ↗Not briefed |
bluesky bluesky-ai | 7d | The entitlement is insane. I've seen another proud slopper crowing about sneaking a Claude-made PR into the Github repo of some anti-AI uti… | The entitlement is insane. I've seen another proud slopper crowing about sneaking a Claude-made PR into the Github repo of some anti-A… | ↗Not briefed |
arxiv arxiv-ai | 7d | Benchmarks Are Surprisingly Compaction-Friendly🔧 GPT-5 | Long horizon agentic tasks are now pushing LLM agents past their context window. To continue the task, the agent scaffold has to free… | ↗Not briefed |
rss techmeme | 7d | US officials say Trump asked Elon Musk to lift Starshield restrictions; Ukraine officials said it was not fully fixed Thursday but was working at higher speeds (Wall Street Journal) | Wall Street Journal: US officials say Trump asked Elon Musk to lift Starshield restrictions; Ukraine officials said it was not fully f… | ↗Not briefed |
rss theverge-ai | 7d | In Meta’s version of the future, I’ll have my Muse call your Muse | Mark Zuckerberg is barrelling toward a future in which an autonomous network of AI agents will run our lives - shopping for us, workin… | ↗Not briefed |
bluesky bluesky-ai | 7d | Technocracy Oligarch threats against government www.vietnam.vn/en/ti-phu-mu... #ai #llm #politics #technocracy #oligarch #power #grab #th… | Technocracy Oligarch threats against government www.vietnam.vn/en/ti-phu-mu... #ai #llm #politics #technocracy #oligarch #power #gra… | ↗Not briefed |
hn hn-ai | 7d | Tell HN: Claude resets expire at an arbitrary time during deadline day🔧 OpenRouter | I was saving a reset on Claude until today and maximising my usage because of it, i.e. I'm almost out of tokens.<p>When I went to… | ↗Not briefed |
hn hn-ai | 7d | Security Sweet: Get your AI agents on a leash for security patch | — | ↗Not briefed |
bluesky bluesky-ai | 7d | I fucking hate Google I fucking hate Microsoft I fucking hate OpenAI, Anthropic, Grok, Gemini, Muse; GET YOUR PLANET-DESTROYING MASS SURV… | I fucking hate Google I fucking hate Microsoft I fucking hate OpenAI, Anthropic, Grok, Gemini, Muse; GET YOUR PLANET-DESTROYING MASS… | ↗Not briefed |
rss simon-willison | 7d | Donating Agent Skills to the Agentic AI Foundation | — | ↗Not briefed |
bluesky bluesky-ai | 7d | Anthropic julkaisi täysin automaattisesti töitä tekevän sihteerin Claude Duet oppii käyttäjän työtehtävät ja osaa pyöriä taustalla ympäri … | Anthropic julkaisi täysin automaattisesti töitä tekevän sihteerin Claude Duet oppii käyttäjän työtehtävät ja osaa pyöriä taustalla ym… | ↗Not briefed |
arxiv arxiv-ai | 7d | Efficient Omni-modal Large Language Model Inference via Per-Query Modality Selection🔧 Modal | Omni-modal large language models (Omni-LLMs) can jointly process a broad range of modalities, such as text, audio, and video, enabling… | ↗Not briefed |
bluesky bluesky-ai | 7d | The losses are gigantic and I expect Anthropic to go bust within a few years but I don't see a reason to make up numbers | The losses are gigantic and I expect Anthropic to go bust within a few years but I don't see a reason to make up numbers | ↗Not briefed |
bluesky bluesky-ai | 7d | The LLM aficionados who need the latest frontier model, and claim local models need 128 Gb RAM (at least!) are just like hifi audiophiles: … | The LLM aficionados who need the latest frontier model, and claim local models need 128 Gb RAM (at least!) are just like hifi audiophi… | ↗Not briefed |
bluesky bluesky-ai | 7d | Here's why the PHP bugs are so dangerous and have created a cyber emergency. | Here's why the PHP bugs are so dangerous and have created a cyber emergency. | ↗Not briefed |
bluesky bluesky-ai | 7d | NEW by me -- The internet's hidden security nightmare — and how AI exposed it --A look at why the PHP bugs are so dangerous and have creat… | NEW by me -- The internet's hidden security nightmare — and how AI exposed it --A look at why the PHP bugs are so dangerous and have… | ↗Not briefed |
hn hn-ai | 7d | Show HN: agent-usage-guard – a Claude Code mod to protect your usage window | — | ↗Not briefed |
hn hn-ai | 7d | Donating Agent Skills to the Agentic AI Foundation | — | ↗Not briefed |